Using the theory of planned behaviour to predict farmer's intention to report livestock depredation and kill hyena

Understanding and managing conservation conflicts is important for stakeholders (e.g., policymakers and practitioners) trying to minimise negative impacts on people and biodiversity. A key component of Namibia’s community-based natural resource management system, besides enabling communities to derive benefits from wildlife, is the monitoring of wildlife and reporting of negative wildlife impacts on human lives and livelihoods. Farmers across Namibia may legally kill carnivores found attacking their livestock and may receive financial compensation if reported within twenty-four hours. Both interventions are intended to offset costs and build tolerance towards wildlife. Expanding the Theory of Planned Behaviour by incorporating Descriptive Norm, we investigated farmers’ Behavioural Intention to (1) legally kill brown Hyaena brunnea and spotted Crocuta crocuta hyena when found killing their livestock, and (2) report livestock depredation incidents to the relevant authorities in two governance contexts – inside versus outside communal conservancies. We hypothesised famers inside communal conservancies would have lower behavioural intentions to kill hyaena and stronger intentions to report livestock depredation compared to farmers outside conservancies. Questionnaire data were collected from 1,139 farmers from inside (n=945) and outside (n=188) communal conservancies. Most respondents reported no intention to kill hyena that killed their cattle, with no significant difference between farmers living inside (89%) and outside (90%) conservancies. Intention to report depredation incidents differed significantly between groups, with 90% of respondents inside conservancies intending to report compared to 78% outside conservancies. Inside conservancies, Attitude was the strongest predictors of farmers’ Behavioural Intention to kill hyena and report incidents of livestock depredation. Outside conservancies, intention to kill hyaena was most strongly associated with Perceived Behavioural Control, while Attitude was the strongest predictor of intention to report. Including Descriptive Norm improved model fit. Our findings highlight how socio-psychological factors differ between governance contexts and how they subsequently influence farmer’s behavioural intentions. Our improved understanding of perceptions underpinning farmers’ decision-making can inform the design of interventions to reduce retaliatory killing and improve reporting of wildlife impacts. Results from this study could also improve the interpretation of national depredation databases and guide more effective mitigation strategies.

Methods
The survey was piloted in early 2019 in five conservancies to identify ambiguities, assess question flow, and improve clarity before full implementation. Between 2019 and 2020, we carried out 1,139 surveys using a questionnaire composed of three sections. The first gathered data on perceived risks and impacts of wildlife on households and characterised farms found within and outside of communal conservancies (Table S2). The first section also gathered information on farmer demographics (e.g., age, education level and household size) and number of livestock owned (Table S3). The final two sections measured Descriptive Norm and four components of the TPB (Attitude, Subjective Norm, Perceived Behavioural Control and Behavioural Intention) enumerating respondents’ perspectives with respect to two distinct behaviours, killing hyena and reporting a livestock depredation incident by hyena. Answers were reported using a 5-point Likert-type scale ranging from 1 = strongly disagree to 5 = strongly agree (Table 1). Respondents answered TPB questions from their own personal view but answered other questions (e.g., number of cattle owned) at the household level. Respondents were categorised as owning livestock or not, and as members of a communal conservancy or not. Respondents’ livestock ownership and conservancy membership were recorded using binary (Yes/No) categories. Table 1 shows constructs and items used to model farmer’s intention to kill hyena that were killing their livestock, and report incidents of livestock depredation, as well as reliability and descriptive statistics of items.

Given the rural setting, most homesteads or villages had low population densities, were often difficult to access, scattered and, in some cases, communities inside communal conservancies were semi-nomadic. For these reasons, we used a combination of snowball and convenience sampling (Newing, 2010), with the household defined as the sampling unit (FAO, 2010). We surveyed either the household owner or another adult member (≥18 years) who knew about the households’ farming practices and wildlife interactions. We aimed to sample a minimum of 40 households per communal conservancy and 40 households per selected area adjacent to communal conservancies (from here on collectively referred to as areas “outside conservancies”), although the final number varied per site depending on respondent willingness and accessibility. Surveys were conducted in English, with the support of local interpreters. In communal conservancies, office staff and community game guards acted as guides and interpreters, while outside conservancies, community members assisted in this role. Interpreters helped translate questions and participants’ responses into the relevant local languages, ensuring clarity and comprehension. Each survey took approximately one hour to administer. Permission to undertake research at each site was obtained from the relevant traditional authorities and conservancy office staff. Data from the questionnaire were collected and were data recorded using ‘Device Magic’ (Device Magic, 2021).